Gemma-SEA-LION-Guard-v2-E2B
Gemma-SEA-LION-Guard-v2-E2B adapts Gemma 4 E2B IT for structured moderation and policy decisions. It supports rules supplied by the application, checks across conversation turns and evaluation of tool calls and responses.
How this release works
The model estimates content-category probabilities and separate probabilities for rule applicability and violation. Its eight-language set includes Burmese, English, Indonesian, Filipino, Malay, Tamil, Thai and Vietnamese. The card supplies a self-hosted decision-server example.Gemma-SEA-LION-Guard-v2-E2B official model card
How to assess it for your work
Practical guidance. Use this member as a compact baseline for a moderation comparison. Include code-switching, quoted harmful content and benign uses of sensitive words. Check whether a decision changes when earlier context is supplied. Keep the policy text precise so a model error is not confused with an ambiguous rule.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Access | API, Downloadable |
| Model identifiers | aisingapore/Gemma-SEA-LION-Guard-v2-E2B |
| Base model | google/gemma-4-E2B-itGemma-SEA-LION-Guard-v2-E2B official model card |
| Context stated by the card | 128K tokensGemma-SEA-LION-Guard-v2-E2B official model card |
| License declared by the model card | MITGemma-SEA-LION-Guard-v2-E2B official model card |
| Access described by the card | Downloadable weights and managed APIGemma-SEA-LION-Guard-v2-E2B official model card |
Related models
- Gemma-SEA-LION-Guard-v2-31B
AI Singapore · SEA-LION Guard v2
A Gemma 4 31B moderation model that scores conversations and tool actions against application-defined rules.
Launch date not confirmedAPI · Downloadable - Gemma-SEA-LION-Guard-v2-E4B
AI Singapore · SEA-LION Guard v2
A Gemma 4 E4B moderation model for custom policy checks across conversations and tool interactions.
Launch date not confirmedAPI · Downloadable